LDD3源码分析之并发与竞态

本文分析LDD3第六章中关于简单休眠的示例代码sleepy.c。

首先列出sleepy.c的完整代码：

#include <linux/module.h>
#include <linux/init.h>

#include <linux/sched.h>  /* current and everything */
#include <linux/kernel.h> /* printk() */
#include <linux/fs.h>     /* everything... */
#include <linux/types.h>  /* size_t */
#include <linux/wait.h>

MODULE_LICENSE("Dual BSD/GPL");

static int sleepy_major = 0;

static DECLARE_WAIT_QUEUE_HEAD(wq);
static int flag = 0;

ssize_t sleepy_read (struct file *filp, char __user *buf, size_t count, loff_t *pos)
{
    printk(KERN_DEBUG "process %i (%s) going to sleep\n",
            current->pid, current->comm);
    wait_event_interruptible(wq, flag != 0);
    flag = 0;
    printk(KERN_DEBUG "awoken %i (%s)\n", current->pid, current->comm);
    return 0; /* EOF */
}

ssize_t sleepy_write (struct file *filp, const char __user *buf, size_t count,
        loff_t *pos)
{
    printk(KERN_DEBUG "process %i (%s) awakening the readers...\n",
            current->pid, current->comm);
    flag = 1;
    wake_up_interruptible(&wq);
    return count; /* succeed, to avoid retrial */
}


struct file_operations sleepy_fops = {
    .owner = THIS_MODULE,
    .read =  sleepy_read,
    .write = sleepy_write,
};


int sleepy_init(void)
{
    int result;

    /*
     * Register your major, and accept a dynamic number
     */
    result = register_chrdev(sleepy_major, "sleepy", &sleepy_fops);
    if (result < 0)
        return result;
    if (sleepy_major == 0)
        sleepy_major = result; /* dynamic */
    return 0;
}

void sleepy_cleanup(void)
{
    unregister_chrdev(sleepy_major, "sleepy");
}

module_init(sleepy_init);
module_exit(sleepy_cleanup);

在模块初始化函数中，注册字符设备”sleepy”时，指定了该设备的读写函数分别是sleepy_read和sleepy_write。当某个进程对sleepy执行读操作时，会进入休眠。当某个进程对sleepy执行写操作时，会唤醒相应等待队列中的所有休眠进程。

为了管理休眠进程，需要建立等待队列，等待队列就是一个进程链表，其中包含等待某个特定事件的所有进程。等待队列通过“等待队列头”来管理，等待队列头是一个类型为wait_queue_head_t的结构体。可以静态初始化一个等待队列头：

DECLARE_WAIT_QUEUE_HEAD(name);

也可以动态初始化一个等待队列头：

wait_queue_head_t my_queue;

init_waitqueue_head(&my_queue);

一个进程要进入休眠，最常用的函数是：

wait_event_interruptible(queue, condition);

queue是等待队列头，condition是一个条件表达式，进程进入休眠前和被唤醒后，都会检查condition的值是否为真，如果不为真，则进程会进入休眠。

对应wait_event_interruptible的唤醒函数是：

wake_up_interruptible(wait_queue_head_t *queue)

sleepy.c第14行定义了等待队列头wq：

14  static DECLARE_WAIT_QUEUE_HEAD(wq);

在sleepy_read函数中，21行调用wait_event_interruptible(wq, flag != 0)进入休眠。所以只要有进程对sleepy执行读操作，就会进入休眠。

在sleepy_write函数中，33行将flag设置为1，然后调用wake_up_interruptible(&wq)将等待在wq上的进程唤醒。

注意，因为在sleepy_read函数中，休眠进程被唤醒后，会把flag重新设置为0，所以虽然全部休眠进程都会被唤醒，但一次只有一个进程能真正继续执行，其它进程会重新休眠。但是为简单起见，这里没考虑并发处理等问题。

另外还可以调用带计时功能的版本：

ssize_t sleepy_read (struct file *filp, char __user *buf, size_t count, loff_t *pos)
{
    printk(KERN_DEBUG "process %i (%s) going to sleep\n",
            current->pid, current->comm);
    //wait_event_interruptible(wq, flag != 0);
    wait_event_interruptible_timeout(wq, flag!=0, 3*HZ);
    flag = 0;
    printk(KERN_DEBUG "awoken %i (%s)\n", current->pid, current->comm);
    return 0; /* EOF */
}

ssize_t sleepy_write (struct file *filp, const char __user *buf, size_t count,
        loff_t *pos)
{
    printk(KERN_DEBUG "process %i (%s) awakening the readers...\n",
            current->pid, current->comm);
    flag = 1;
    wake_up_interruptible(&wq);
    return count; /* succeed, to avoid retrial */
}

要测试sleepy模块，可以先创建sleepy_load和sleepy_unload脚本。

sleepy_load脚本的内容如下：

#!/bin/sh
# $Id: complete_load,v 1.4 2004/11/03 06:19:49 rubini Exp $
module="sleepy"
device="sleepy"
mode="666"

# Group: since distributions do it differently, look for wheel or use staff
if grep -q '^staff:' /etc/group; then
    group="staff"
else
    group="wheel"
fi

# invoke insmod with all arguments we got
# and use a pathname, as insmod doesn't look in . by default
/sbin/insmod ./$module.ko $* || exit 1

# retrieve major number
major=$(awk "\$2==\"$module\" {print \$1}" /proc/devices)

# Remove stale nodes and replace them, then give gid and perms
# Usually the script is shorter, it's scull that has several devices in it.

rm -f /dev/${device}
mknod /dev/${device} c $major 0

chgrp $group /dev/${device}
chmod $mode  /dev/${device}

sleepy_unload脚本的内容如下：

#!/bin/sh
module="sleepy"
device="sleepy"

# invoke rmmod with all arguments we got
/sbin/rmmod $module $* || exit 1

# Remove stale nodes

rm -f /dev/${device}

本文参考： http://blog.csdn.net/liuhaoyutz/article/details/7388163